In a 22-month period, strains of Staphylococcus aureus resistant to methicillin and multiple aminoglycosides, (designated MARS) were recovered from 108 inpatients with nosocomial infections at a hospital in the midwestern United States. Sixty-six of these patients were staying in a burn unit, and 42 were on other hospital wards. Among the patients with burns, MARS were recovered from the burn wounds of 64%; 32% of the patients with burns had MARS bacteremia. The patients without burns were age-matched with patients with nosocomial infections caused by antibiotic-susceptible strains of S. aureus. Patients from whom MARS were isolated had a longer mean hospital stay (79.6 days vs. 36.9 days; P less than 0.01), developed infection later (26.5 days vs. 13.5 days after admission; P less than 0.01), and had received antibiotic therapy before infection more often (81% vs. 38% of patients; P less than 0.01) than patients in the comparative population. Types of infection and incidences of death and bacteremia were similar in the two groups. Antibiotic-resistant strains of S. aureus may cause serious infections and significant mortality.

We have studied the effect of age at diagnosis in the proband and parental status [normal versus non-insulin-dependent diabetes (NIDD)] on the cumulative risk to age 40 (CR40) for insulin-dependent diabetes (IDD) in sibs of IDD probands in 493 families. We found a significantly increased cumulative risk to age 40 fro IDD in sibs of probands with disease diagnosed before age 10 (8.5 +/- 2.0%) as compared with that in sibs of probands with disease diagnosed after age 10 (4.6 +/- 0.8%) (X2 = 7.6, P = 0.006). Within the family subset with probands of earlier IDD onset (before age 10) we also found an increased CR40 for IDD in the sibs of the probands in families with an NIDD parent (with NIDD parent: CR40 = 7..5 +/- 2.0%, X2 = 12.8, P less than 0.0005). These data are compatible with the theory of heterogeneity (genetic and/or environmental) of IDD and a possible relationship between IDD and NIDD.

A B S T R A C T INTRODUCTIONOne of the few seemingly well-established facts about the etiology of insulin-dependent diabetes mellitus (IDDM)l is that extensive heterogeneity is present (1-8). Thus, several genetic mechanisms are likely to be involved. These may possibly range from multigenic (with each locus contributing only a small effect) to the major genetic effect of Mendelian modes of inheritance. The recent reports of association between the histocompatibility system (HLA) and IDDM (5,6) indicate a direct and important effect of the HLA antigens on the risk of the disease. One possible explanation of this association is linkage between the two traits; we have conducted a series of linkage analysis studies to examine these possibilities. Genetic studies of IDDM must be designed to answer a specific research question under circumstances that minimize the "noise" of heterogeneity. We have, therefore, sampled multiplex families (i.e., those families with